Johan,
> AFAICS all synchronous calls are compatibility wrappers and thus
> deprecated.
I'm only trying to remove usb_request_irq(), usb_release_irq(),
usb_request_bulk(), and usb_terminate_bulk().
We'll keep usb_control_msg() and usb_bulk_msg() even though
they are synchronous.
> Is synchronous against policy or is it because it usually
> isn't usefull ?
Not against policy. You just can't get any kind of decent
streaming throughput with it, so drivers should use the async.
URB-style requests in cases where performance is important.
For some cases, usb_bulk_msg() might be OK.
usb_control_msg() is generally for config/setup/init type
stuff, so it's not a performance area (although a few devices
use control messages for status reporting or as a generic
read_register/write_register interface, where performance
is more critical, so they should consider using a Control-type
URB for this.)
> The more I think about it, synchronous indeed isn't nescessary.
> The reason for the synchronous calls is that the Speedtouch contains a
> statemachine that follows the state of the ADSL line. At the
> moment, it
> is implemented as a thread. The more I think about it, the
> more i feel
> it doesn`t belong there and I am strongly considering to move it to
> userspace. This would make configuration a lot easier too.
> Opinions ?
Using usbdevfs or devfs or something else?
~Randy
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
